Comment by rickcarlino

Neat idea! You might be able to deduplicate the list by converting the decisions to embeddings and running k-means clustering.